Fix a case where zmq_poll() (poll-based version) could go off into a busy-loop
if no revents are returned and the timeout passed in is zero.

* maint:
  zmq_msg_move called on uninitialised message in xrep_t::xrecv -- fixed

 * Assertion to check that pgm_getaddrinfo is actually returning something.
 * Missing pgm_connect call.
 * Typo on TOS causing immediate abort.
 * Placeholder calls for timeouts whilst continuing spin loop functionality.
 * OpenPGM v5 now supports reference counting so remove init checks.
 * Duplicate UDP unicast port setting, requires one unicast and one multicast.
 * Incorrectly set socket rcvbuf size with sndbuf.
 * Replace std::lexicographical_compare of TSI's with long word integer comparisons.
 * pgm_socket_t::receive returns -1 on no data.

* maint:
  crash when closing an ypipe -- fixed

* maint:
  if TSC jumps backwards (in case of migration to a different CPU core) latency peak may occur -- fixed

Conflicts:
	src/app_thread.cpp

if TSC jumps backwards (in case of migration to a different CPU core) latency peak may occur -- fixed

The flag in the socket has clear semantics now -- it tracks whether corresponding context was closed, it doesn't track whether zmq_close was called on the socket itself

termination of pipe via delimiter message could stuck when no data were read from the pipe (because connection wasn't active) -- fixed